https://issues.apache.org/bugzilla/show_bug.cgi?id=43925
--- Comment #15 from Issa Gorissen <flo...@usa.net> --- I made some changes to the test case you provided to excluded the random (so that the random numbers are always the same for both cases) and run the test for one target only. JVM is Java HotSpot(TM) 64-Bit Server VM (25.5-b02) for linux-amd64 JRE (1.8.0_05-b13), ran with settings -server -Xloggc:gc.log -Xmx4G. I don't get the same results as yours and GC is running a lot more on your class than Tomcat's. (I can send you the results if you'd like) It's true I can increase the number of iteration with your class before getting a OOE more than with Tomcat's. But is this test case emulating real usage ? My next tests will be to run both class under Tomcat with a test case of several pages with tags triggered by jmeter. This will take me some time so I will not be back before several weeks. Thx -- You are receiving this mail because: You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
